The report criticizes the AI safety debate for being too focused on catastrophic risk, arguing that it distracts from more immediate issues. It calls for mandatory safety tests for high-risk models and suggests that concerns about existential risk are exaggerated. The report also highlights the need to address the ease with which misinformation can be created and spread, and the use of copyrighted material to train LLMs. It recommends that the government's AI Safety Institute should carry out an assessment of engineering pathways to catastrophic risk and warning indicators.
Key takeaways:
- The U.K. government is being criticized for taking a too “narrow” view of AI safety and risks falling behind in the AI goldrush, according to a report by the parliamentary House of Lords’ Communications and Digital Committee.
- The report suggests that the government should focus more on near-term security and societal risks posed by large language models (LLMs) such as copyright infringement and misinformation, rather than hypothetical existential threats.
- The report also calls for the Government to make market competition an “explicit AI policy objective” to guard against regulatory capture from incumbents such as OpenAI and Google.
- The report criticizes the government's focus on catastrophic risks, suggesting it is exaggerated and distracts from more immediate issues that LLMs are enabling today, such as the spread of misinformation and copyright infringement.